What We Know about Demand Surge: Brief Summary
Anna H. Olsen1 and Keith A. Porter2
Abstract: Demand surge is a process resulting in a higher cost to repair building damage after large disasters than to repair the same damage
after a small disaster; this higher cost can be an additional 20% or more. It is of interest to insurers, regulators, property owners, and others.
Despite its importance, demand surge has no standard definition or generally accepted predictive theory of its mechanisms and quantitative
effects. By studying the circumstances of natural disasters that did and did not cause demand surge, common explanatory themes emerge from
these historical events that may describe why and how much losses increase in some disasters. The themes are: total amount of repair work;
timing of reconstruction; costs of materials, labor, and equipment; contractor overhead and profit; the general economic situation; insurance
claims handling; and decisions of an insurance company. The development of these themes will aid in constructing a mechanistic, empirically
supported approach to modeling demand surge. DOI: 10.1061/(ASCE)NH.1527-6996.0000028. © 2011 American Society of Civil
Engineers.
CE Database subject headings: Risk management; Insurance; Natural disasters; Buildings; Rehabilitation.
Author keywords: Demand surge; Natural disasters; Damage; Insurance.
Introduction
Demand surge is an issue for individuals and institutions that sustain losses in large-scale natural disasters, particularly for property
insurers and governments that finance reconstruction. Estimates of
demand surge following large-scale natural disasters have quantified a general increase of costs ranging from 10 to 40% following
Hurricane Katrina (Guy Carpenter 2005) to 50% after Cyclone
Larry [Australian Securities and Investments Commission (ASIC)
2007]. For specific materials and labor items, news reports have
documented price increases of 30% for oriented strand board following Hurricane Katrina (Grogan and Angelo 2005) to a 2,000%
increase for securing a tarpaulin to a damaged roof after the 1999
Sydney hailstorm (Sweetman and Morris 1999). The higher repair
costs at each property result in a greater loss for an insurer that
indemnifies many properties in an affected area. For a single
insurer, this additional loss caused by demand surge may mean
the difference between survival and ruin. For example, 20th
Century Insurance, based in the Los Angeles area, was nearly bankrupted by claims following the 1994 Northridge Earthquake
(Stavro 1998), a disaster that produced a reported 20% demand
surge (Kuzak and Larsen 2005).
Commercial catastrophe modelers, such as Applied Insurance
Research (AIR), EQECAT, and Risk Management Solutions
(RMS), develop models of demand surge. One of the writers
(Porter) created EQECAT’s first demand-surge model in the
mid-1990s, which is approximately when RMS and AIR first
developed theirs. The catastrophe modelers describe their models
1
Willis Research Fellow and Postdoctoral Scholar, Univ. of Colorado at
Boulder, Boulder, CO 80309 (corresponding author). E-mail: anna
[email protected]
2
Associate Research Professor, Univ. of Colorado at Boulder, Boulder,
CO 80309.
Note. This manuscript was submitted on January 28, 2010; approved on
May 19, 2010; published online on April 15, 2011. Discussion period open
until October 1, 2011; separate discussions must be submitted for
individual papers. This paper is part of the Natural Hazards Review,
Vol. 12, No. 2, May 1, 2011. ©ASCE, ISSN 1527-6988/2011/2-62–71/
$25.00.
publicly but keep the details private as intellectual property. As a
result, there seems to be no independent, public examination of the
commercial demand-surge models to test their methods and duplicate their results. The Florida Commission on Hurricane Loss
Projection Methodology does assess commercial catastrophe models to approve their use for rate filings in Florida, but the modelers’
methodologies remain confidential (Florida Statute 2009). Because
these models are proprietary, there is no synergy of the modelers’
insights into demand surge.
Demand surge bedevils consumers and those government agencies that serve a consumer-protection role. Since proprietary models are somewhat opaque, skeptical insurance consumers and their
advocates have an ipso facto license to question the validity of
demand-surge models. Consumer advocates have suggested the
possibility that additional costs attributed to demand surge are
illusory or perhaps can be controlled by the insurer (Ruquet
2009). An insurance company may counter that, to be economically
viable, it must use the best available model to anticipate any
demand-surge costs and reflect these costs in policy premiums.
These extra premiums become a source of conflict between insurers, policyholders, and consumer advocates.
This paper develops an understanding of demand surge as a
socioeconomic phenomenon associated with large-scale natural
disasters. It provides evidence of demand surge as the outpacing
of supplies (of reconstruction materials, labor, equipment, financing, or some combination of these) by their respective demands
after a natural disaster. This introduction has described the problem.
Subsequent sections will discuss some terminology and definitions, followed by descriptions of existing models of insured and
economic loss after natural disasters, with particular attention to
demand surge. Finally, some common themes of demand overwhelming supply will be distilled from observations of historical
natural disasters. These themes provide possible explanations for
the mechanics of demand surge, and they inform a quantitative
model the writers are now developing. This work does not fully
and completely describe and explain the phenomenon of demand
surge. Rather, it summarizes the current state of knowledge and
collects qualitative evidence for demand surge from historical
events. This paper condenses a longer report (Olsen and Porter
62 / NATURAL HAZARDS REVIEW © ASCE / MAY 2011
Downloaded 28 Jun 2011 to 198.11.27.29. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
2010), in which the reader can find a more complete discussion of
the issues presented in this paper.
Terminology
“Demand surge” has various, imprecise definitions, and the term
has various, inconsistent connotations. The basic understanding
of demand surge is exemplified by the definition of the Actuarial
Standards Board: “A sudden and usually temporary increase in the
cost of materials, services, and labor due to the increased demand
for them following a catastrophe” (Subcommittee on Ratemaking
of the Casualty Committee 2000). The five parts of this definition
can be made more explicit: (1) reconstruction materials prices,
labor wages, and the costs of reconstruction services in general,
increase (2) because of a significant demand for reconstruction
activities (3) soon, if not immediately, after (4) a large-scale natural
disaster, and (5) these cost increases remain unusually high for a
limited period of time before returning to a lower level once supply
satisfies demand. Note that, in this definition, demand surge is the
increase of costs resulting from increased demand; it is unclear
whether demand surge also encompasses the increased demand
for quantities of materials, labor, and services. In other words, there
is no clear distinction between the underlying phenomenon and the
metric used to measure the phenomenon.
Other definitions and usages of “demand surge” limit the scope
of the term. Consider the following two quotations from the insurance industry literature: “With demand surge, insured losses creep
upward due to the increased price of construction materials and
labor following large losses such as Katrina and Rita” (Howard
2005); and “Major catastrophes, such as earthquakes, hurricanes,
and wildfires can often create a demand surge for materials and
labor, resulting in increased costs to replace damaged property”
(Federal Alliance for Safe Homes and The Actuarial Foundation
2006). Here, demand surge refers specifically to temporary increases in materials prices and labor wages following large-scale
natural disasters. It does not refer to other cost increases, such
as higher rates of contractor overhead and profit or waivers of
multiple insurance deductibles in clustered events, even if these
cost increases can be explained as demand for some product or service overwhelming its supply.
Furthermore, there is no consensus on what specific material
and labor costs contribute to demand surge. Labor costs are the
wages paid to workers in the construction industry. However, to
be amenable to study, one must still determine precisely which
workers, in what location, and over what time period these should
be considered, as well as how to characterize this information in a
model. Material costs probably refer to the retail or wholesale
prices charged by or to the construction industry for the materials
required to repair the damage. Again however, the geographic and
temporal scope of the products, and which ones, must be crisply
defined before it is practical to test any hypotheses about their
influence on demand surge.
Although limiting a demand-surge definition to increases in
materials prices and labor wages helps to clarify the definition,
there is no evidence that these are the only two—or even the most
important two—drivers of demand surge. At present, there is no
apparent reason to limit the definition to increased materials prices
and labor wages when there are other reconstruction resources that
may also be in short supply after large-scale natural disasters, such
as equipment, financing, and the number of construction contracting firms, which can be the primary provider of labor at
reconstruction sites.
“Demand surge” has also been used to explain the difference
between an expected, or modeled, loss and the realized, or actual,
loss in a large-scale natural disaster. For instance:
After the 2004 hurricanes in Florida, an attempt was made to
explain the often underestimated losses by citing such effects
as demand surge and major catastrophe surcharges. Following
Katrina, many insurers found that the losses they finally had to
pay were often far higher than their initial forecast. This was
due to the fact that the scale of a major catastrophe is
enhanced by the shortage of resources (construction materials
and workers needed for reconstruction work) and the limited availability of infrastructure installations (Munich Re
Group 2006).
Using “demand surge” to refer to unexpected losses can obscure
alternate explanations for the difference between expected and realized losses. Examples include modeling error (the inherent discrepancy from the use of a model to estimate loss instead of summing
actual losses), input error (the discrepancy between an estimate of
hazard, vulnerability, and exposure and the true values of these
model inputs), and error in measuring actual loss (which is commonly extrapolated from losses reported by a subset of insurers).
Using “demand surge” as an underlying but unquantified cause of
the observed discrepancy raises the question: is demand surge an
observable, model-independent, real-world phenomenon, or is it a
product of the conceptualization and quantification of losses from
natural disasters? Is demand surge a carpet under which we sweep
the dust of modeling error?
Most observers’ use of “demand surge” is quite general and similar to the definition of the Actuarial Standards Board presented
previously. The limited sense of demand surge, as an increase in
material and labor costs, is often the first and primary description
of increased losses after large-scale natural disasters. But additional
explanations are offered that augment the limited definition. For
example:
With a major windstorm, overall losses increase not only
because of economic shortages in roofing materials and
workmen, but because insurance companies are unable to
police the volume of claims. [Called “loss-cost inflation” or
“demand surge.”] For the largest of all catastrophic losses,
politicians put pressure on insurance companies to relax their
internal rules and fast track the processing and repayment of
claims (Risk Management Solutions, Inc. 2000).
In this example, the quote offers increasing material and labor
costs first to explain larger losses in large-scale natural disasters and
then provides additional, specific explanations. This type of definition begs the question: what is a reasonably complete list of these
additional explanations? How does someone determine whether
each explanation contributes to demand surge or to a different
phenomenon, such as business interruption, in which repair delays
or shortages of alternative facilities inhibit return to normal
operations?
Several general definitions or usages of “demand surge” add
“services” to the short list of labor and materials to explain higher
costs after large-scale natural disasters, but the reference to services
is unclear. For example, precisely which services ought to be considered in the following quote: “Demand surge [is] the increase in
costs of materials, services, and labor due to increased demand following a catastrophic event” (AIR Worldwide Corporation 2007)?
“Services” might refer to expenditures by insurers to claims adjusting firms, or by insurers to construction contracting firms (over and
above the contractors’ labor and material costs), or by construction
NATURAL HAZARDS REVIEW © ASCE / MAY 2011 / 63
Downloaded 28 Jun 2011 to 198.11.27.29. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
contracting firms to businesses such as hotels, restaurants, and
equipment-rental companies that serve them while they mobilize
outside their home market, market their services, or do the repairs.
Since insurance claims include additional living expenses and
other time-element losses, and since restoring operations could
involve purchases from great distances for company-specific needs,
services in this context could mean virtually any expenditure paid
to businesses at any distance from the disaster by any insured entity
during the life of any time-element claim. These examples indicate
that the current understanding of demand surge is not sufficiently
developed to easily and routinely specify the geographic, temporal,
or sectoral scope of services that contribute to increased losses
following large-scale natural disasters.
In addition to “demand surge,” there are other terms that refer to
the same phenomenon or similar ideas. Two groups have separately
offered distinct terminology to identify and explain the increased
losses after large-scale natural disasters. Risk Management
Solutions proposed “loss amplification” as the increase in loss following large-scale natural disasters that is not adequately modeled
by assuming independent losses at exposed properties (Grossi and
Muir-Wood 2006). RMS identifies four issues that contribute to
loss amplification: economic demand surge (“increases in the costs
of building materials and hourly rates for labor, as demand exceeds
supply”); repair delay inflation (“increases in the amount of damage or costs of interrupted business associated with delays in making the repairs”); claims inflation (“increases in the size of claims as
the ability of insurance adjusters to inspect properties is impeded
due to the number of claims”); and coverage expansion (“the degree
to which the terms of the original insurance contract becomes
expanded to cover additional sources of loss or higher limits”).
The same concepts are presented elsewhere by RMS with slightly
different names.
Risk Frontiers, formerly the Natural Hazards Research Centre,
studies topics of interest to the insurance industry and used the term
“postevent claims inflation” to refer to the increase of payments for
insurance claims after major disasters (McAneney 2007). They use
“demand surge” to refer specifically to the outpacing of supply by
the demand for goods and services, which results in higher prices.
These terminologies and the accompanying explanations help to
define the idea of demand surge, but none provides a precise definition of the phenomena. The same challenges of circumscribing
the scope of these terms exists as when this phenomenon is called
demand surge.
The writers intentionally do not offer new terminology or a precise definition of demand surge. This work collects and distills
anecdotal evidence for demand surge in historic events and proposes hypotheses to explain the phenomenon. Since this paper does
not provide—nor have the writers found in published studies—
clear evidence for what demand surge is and is not, it seems premature to introduce additional terminology. The existing terminology, flawed as it may be, is sufficient while researchers develop a
sound understanding of reconstruction costs after large-scale natural disasters. The writers expect that the knowledge from future
work will suggest improved terminologies and precise definitions
of the mechanisms that drive the complex phenomenon we currently call “demand surge.”
Current Demand-Surge Models
Models to describe how and why losses increase as a result of a
large-scale natural disaster have been developed in the last two decades. Current demand-surge models capture parts of the phenomenon, but no published model fully describes how much and why
losses increase as a result of the disequilibrium of demand and supply after large-scale natural disasters. Perhaps there has not yet been
enough accumulated knowledge of demand surge to develop and
test such a comprehensive model. This section describes a sample
of the limited documentation of demand-surge models by commercial catastrophe modelers and by professional economists.
Commercial Catastrophe Models
Broadly defined, a catastrophe model is a tool that provides a quantitative estimate of risk to a single property or a collection of properties, usually from natural hazards, and most often for estimating
insurance claims for a particular insurance company or reinsurance
company in current or future natural disasters. Commercial catastrophe models typically apply a series of stochastic events (for
example, earthquakes in California, flooding in continental Europe,
cyclones in Australia) to estimate the environmental excitation
(shaking intensity, flood depth, windspeed) imposed on a mathematical model of the properties. Then, the model relates intensity
of excitation to repair cost normalized by replacement cost (the
relationship is called a vulnerability function) to produce an estimate of the damage factor at each property. Multiplying the damage
factor by the estimated replacement cost produces an estimate of
total repair cost. Applying deductibles and limits to each policy
and summing over policies produces an estimate of the total payments the insurer would make to settle claims. Uncertainties in each
step are propagated, either through simple Monte Carlo simulation
or more-sophisticated variants, to produce a probabilistic estimate
of loss, either for single-event scenarios or for suites of possible
future events (for example, catalogs of earthquakes).
Building owners routinely use these models to decide how much
insurance to buy; insurers use them to help decide what to charge
for insurance and how much reinsurance they need; and reinsurers
use the results to help them price reinsurance and ensure a reasonable probability of survival. See Grossi and Kunreuther (2005) for a
more extensive discussion.
As frequently employed, the demand-surge component of a
catastrophe model increases the calculated ground-up loss for a
property. (The ground-up loss is the monetary cost to repair damage
and continue operations, before applying insurance deductibles, copays, or limits.) A demand-surge factor multiplies the ground-up
loss, either at the individual-property level or the portfolio level,
by a factor, typically between 1.0 and 1.6. Depending on the model,
this factor can be determined on the basis of the expected loss to the
insurance industry as a whole or on the basis of additional information, such as the affected region, peril, and type of property.
Often, the model end-user can change this factor or turn off the
calculation of demand surge as part of the model’s options.
Though unaware of any publication of a commercial catastrophe
modeler’s demand-surge model, one of the writers (Porter) participated in developing one for EQECAT in the mid-1990s. It has since
been enhanced or replaced in some way, but in brief, the early
model worked as follows:
1. Regression analysis was used to relate damage factor to environmental excitation for past claims, resulting in vulnerability
functions;
2. The regression analysis was performed using data from supposedly non-demand-surge-inducing events and again from an
event that supposedly did produce demand surge (Hurricane
Andrew);
3. The ratio of the latter to the former was calculated as a function
of environmental excitation (windspeed) for broad classes of
properties (that is, residential or commercial properties on
the Gulf and Atlantic coasts of the United States); and
64 / NATURAL HAZARDS REVIEW © ASCE / MAY 2011
Downloaded 28 Jun 2011 to 198.11.27.29. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
4. This ratio was applied at the per-property level, along with a
multiplier that reflected the degree to which the event was
expected to produce demand surge at a macroscopic level,
from nil to 100% of the Andrew-level increase.
In the form of an equation, the model was
L¼
X
V i yi ðsi Þ½1 þ Pðsi ÞGðTÞ
ð1Þ
i
where L = ground-up loss at the portfolio level; i = index to individual properties in the portfolio; V i = estimated replacement cost
for property i; yi ðsi Þ = mean damage factor for property i considering its structure type and given that it is exposed to excitation of
intensity si ; Pðsi Þ = maximum effect of demand surge on a property
when it is subjected to excitation si (a property-level effect); and
GðTÞ = degree to which the event as a whole (such as a hurricane)
would produce demand surge, given the estimated societal-level
total loss T. PðsÞ might vary from 0 to 0.6, for example, in certain
property classes where demand surge can reach 60%, with the lowest effect (0) occurring at low excitation (because damage is so low
repairs can wait) or at high excitation (because damage is so severe
that no amount of hurry will make a difference). It achieves its
maximum value at intermediate excitation, where significant repairs are needed and speed is of the essence. G would vary from
0 (for smaller events like Hurricane Hugo in which the required
repairs did not heavily stress available supplies) to 1.0 (for large
ones like Andrew). T was parameterized as total estimated repair
costs normalized by estimated annual construction contracting revenues of businesses within a given radius of the event’s nominal
location, such as hurricane landfall. (This model used a radius
of 160 km for regions with only one major interstate corridor providing access, and 480 km otherwise.) The model attempted to reflect three hypothesized aspects of demand surge: that repair delays
matter only at moderate levels of damage (captured through P); that
the ratio of demand to local supplies matters; and that access matters (the last two issues parameterized through T).
This description leaves out much detail and is limited in too
many ways to catalog in a brief space. However, it gives a sense
of the simplicity of an early demand-surge model and highlights a
few important points: that demand surge has been used to capture
past discrepancies between observation and expectation; that at
least this model encoded the modeler’s unproven hypotheses of
a few major mechanisms of demand surge; and that, while hindcasting past losses, it does not provide much explanatory power.
However, there have been few attempts to determine the accuracy
of these models for loss estimation (Rose 2004).
Hallegatte (2008) explicitly included demand surge within an
input-output model framework. The study assumed that demand
surge is “driven by the unbalance between the large demand in
the reconstruction sector and the insufficient production capacity.”
Commodity prices were assumed to be proportional to the underproduction of the commodity, and wages were assumed fixed when
determining profits in economic sectors (an assumption made to
simplify the model, with acknowledgment that in the real world,
these calculated profits may be redistributed to the workers, to
shareholders and owners, and to future investment). In the simulation of post-Katrina reconstruction in Louisiana, Hallegatte (2008)
found the total reconstruction costs to be US$121 billion as opposed to $107 billion assuming pre-Katrina price levels—a 13%
increase that Hallegatte attributed to demand surge. Although demand surge was included, the purpose of the study was to assess the
change in value added to the Louisiana economy caused by Hurricane Katrina, considering direct and indirect losses.
Florida Public Hurricane Loss Model
The Florida Public Hurricane Loss Model predicts losses caused by
Atlantic hurricanes and includes a demand-surge model. The
Florida Office of Insurance Regulation funds the model, and researchers at the Laboratory for Insurance, Economic, and Financial
Research at Florida International University lead the model’s development. According to the 2009 Florida Commission on Hurricane
Loss Projection Methodology submission, the demand-surge
component of the Florida Public Model applies “weighted average
demand-surge factors” to the loss from each event in the stochastic
set (Florida International University 2009). The model assumes that
demand surge is affected by the insurance coverage [that is, structure, appurtenant structure, contents, or additional living expenses
(ALE)], region of Florida (either “Northeast/North Central,”
“Northwest,” “Central,” “South (except Monroe County),” or
“Monroe County”), and the modeled total statewide loss without
demand surge. The demand-surge factor for structure coverage is
F structure ¼ c þ p1 lnðLstate Þ þ p2
ð2Þ
where c = parameter; p1 = parameter with one value for Monroe
County and another value for all other regions; p2 = parameter with
a different value for each region; and Lstate = modeled total statewide loss (an attempt to capture demand). The demand-surge factor
for appurtenant structure coverage is the same as the factor for
structure coverage. The factor for contents coverage is
Regional Economic Models Used for Demand Surge
Researchers interested in natural disasters have repurposed regional
economic models to study losses in natural disasters. These models
distinguish economic sectors as well as the interactions between
them, both locally and with other geographic regions. They are
most often used to estimate societal losses as a direct or indirect
result of a natural disaster, where “loss” now refers to the difference
in economic output had there been no disaster versus the output
following a disaster.
There are two common, and one less common, types of models
used to estimate macroeconomic losses in natural disasters. The
two major types are input-output and computable general equilibrium models, and the minor type is an econometric model. Each
model has its advantages, disadvantages, and limitations. It is generally believed that estimates from input-output models are an
upper bound, whereas estimates from computable general equilibrium models are a lower bound on economic loss (Okuyama 2007).
F contents ¼ 1 þ 0:3ðF structure 1Þ
ð3Þ
and the factor for ALE is
F ALE ¼ 1:5F structure 0:5
ð4Þ
The researchers used the following data to develop the model’s
functional form and determine values for the parameters: construction-cost indexes for Florida zip codes from Marshall & Swift/
Boeckh (MSB) (for structure coverage); the household furnishings
and operations index of the Consumer Price Index for Miami-Fort
Lauderdale (for contents); and insured losses caused by Hurricanes
Andrew, Charley, and Frances (for ALE). While developing the
model, the researchers inferred what the MSB construction-cost
index in a region affected by a hurricane would have been if no
hurricane had passed, and they assumed that any difference
NATURAL HAZARDS REVIEW © ASCE / MAY 2011 / 65
Downloaded 28 Jun 2011 to 198.11.27.29. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
between this no-hurricane, inferred value and the with-hurricane,
actual value was entirely attributable to demand surge.
Common Themes of Demand Surge
With this background in mind on the nature of the problem and past
work to model demand surge, focus now shifts to available historical evidence that can be used to inform a deeper understanding of
the mechanisms of demand surge. Past events provide insight into
the underlying supply and demand processes that result in higher
reconstruction costs. To impose some structure to the discussion,
the evidence is grouped into broad themes. The word “theme” is
used intentionally to indicate that, in a particular event, the specific
expression may be unique, but nonetheless, there is an overarching
idea common to demand surge in many past events.
Amount of Repair Work
The total amount of repair work in a region partly defines the
“demand” of demand surge. Conceptually, the event causes initial
damage at potentially millions of properties according to the environmental excitation and vulnerability of each property exposed to
loss. Decisions of all stakeholders made before and after the event
affect what repairs are done and in what order.
Anecdotes related by construction contractors, claims adjusters,
and cost estimators with whom the writers spoke suggest that
whether the property is insured strongly affects the amount of repair
work actually accomplished and how quickly the repairs are
effected. For example, the insurer is sometimes required to pay
for repairs that an uninsured property owner might not otherwise
elect to pay for. An example is the “reasonably uniform appearance” standard required by insurance regulations in several states,
such as the statute in Nebraska that reads: “When a loss requires
replacement of items and the replacement items do not reasonably
match in quality, color or size, the insurer shall replace all items in
the area so as to conform to a reasonably uniform appearance. This
applies to both interior and exterior losses. The insured shall not
bear cost over any applicable deductible” (Nebraska Administrative
Code 2009). Thus, in regions where a substantial fraction of
damage is insured, requirements such as reasonably uniform
appearance can substantially raise the total amount of repair work
actually performed at an insured property when compared with an
uninsured property. Thus, the demand for repair services increases,
and costs consequently increase as well.
Individual property owners may attempt to reduce damage to
their properties in disasters of any size. A property owner may take
measures before the event to prevent damage or make emergency
repairs immediately after to reduce additional damage. The scale of
the disaster, however, may affect these efforts. Damage may accumulate in clustered events, possibly negating any efforts of the
property owner. In a catastrophe, properties may be more likely
to decay because repairs are delayed by the large scope of damage.
For example, mold can flourish in hot, water-damaged buildings
when regional electric power is unavailable for an extended period
of time. Thus, owners cannot quickly cool and dry buildings. This
happened after Hurricane Katrina, for example, at Xavier College
in New Orleans, as described in Mosqueda and Porter (2007). In
short, the amount of damage at an individual property may be
greater because the property was damaged in a large-scale natural
disaster.
The level of repair required by building codes, and enforced by
the building authority, may also affect the amount of work at a
property. The building code in force during the reconstruction
period may require a heavily damaged structure to be built to a
higher standard than what was required when it was originally
built. The rebuilding requirements affect the amount and type of
construction materials and the necessary skill of the labor.
Although building codes tend to become more strict over the years,
there can be exceptions; building codes may not be enforced [for
example, before Hurricane Andrew (Sirkin 1995)], or the local
building department may temporarily suspend certain provisions
to allow for speedier recovery [for example, after Hurricanes
Iwa and Iniki (Chock 2005)].
Judgment may also affect the amount and speed of work performed at a damaged property. Contractors and insurance claims
adjusters may be pressured to quickly define the amount of work
to be done. Adjusters might have a long list of properties to visit,
making each loss assessment less thoroughly than they otherwise
would in a smaller event (Thomas 1976). Contractors and claims
adjusters may not have enough, or the right, information available
about repair work at a property at the time of a repair estimate or
claim adjustment. An initial assessment of damage may not identify
all damage, and unanticipated damage may be encountered only
after demolition and repair work have begun (Tim Wilson, personal
communication, July 2008). These types of judgments about the
amount of repair work following a catastrophe must be made
but may not be fully informed. The pressure of a high volume
of repair work seems likely to exacerbate problems of making
quick judgments.
Costs of Materials, Labor, and Equipment
Local supplies of materials, labor, and equipment are typically
employed first to rebuild after natural disasters. When the local
supplies are outstripped by demand, prices rise. Three historical
examples touch on price increases for material, labor, and equipment, over several centuries and two continents. See Olsen and
Porter (2010) for more examples.
An extratropical cyclone devastated southern and central
England on November 26–27, 1703. There was extensive and severe damage to buildings, especially in London and Bristol, contributing to the 480-km-wide damage footprint (Hamblyn 2005).
In the immediate aftermath of the storm, prices of materials rose
significantly. Defoe (1704) reported that the price of plain roofing
tiles rose from 21 shillings per thousand to 6 pounds, a 470% increase. The price of pantiles (commonly used as a roofing tile) rose
from 50 shillings per thousand to 10 pounds, a 300% increase.
Following the immediate rise of prices, they fell. Defoe stated that
this fall was not explained by the meeting of demand for roofing
tiles with adequate supplies. Rather, property owners and tenants
could not pay the heightened costs. The roofs of some properties
went unrepaired, and the occupants were exposed to the elements
that winter. The roofs of other properties were repaired temporarily
with wood planks (“deal boards”) until enough tiles could be manufactured during the tile-making season. Defoe also reported that,
after the windstorm, property owners were “fortifying their Houses
against the Accidents of Weather by Deal Boards, old Tiles, Pieces
of Sail-Cloth, Tarpaulin, and the like.” That is, they were substituting any appropriate and available materials to temporarily, if not
permanently, repair their buildings.
More recently, the earthquake that occurred north of Charleston,
South Carolina, on August 31, 1886, caused extensive destruction
in Charleston and created a demand for labor that far exceeded the
local supply. Wage rates for skilled and unskilled labor increased
dramatically above the preearthquake levels. Seven days after the
earthquake, the Knights of Labor union authorized a modest,
50-cent-per-day increase in union wages (News and Courier
1886b). However, skilled and unskilled laborers were actually
66 / NATURAL HAZARDS REVIEW © ASCE / MAY 2011
Downloaded 28 Jun 2011 to 198.11.27.29. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
commanding wage rates more than double the preearthquake rates.
Union bricklayers would work for no less than US $5 per day, a
67% increase above the preearthquake rate, and the union president
stated that “the bricklayers were entirely satisfied with … $5 a
day” (News and Courier 1886a). There were rumors, however, that
bricklayers were actually receiving $6 or $8 per day, a
100–170% increase (News and Courier 1886c).
After the 1992 Hurricane Andrew in South Florida, heavy
equipment was in demand for debris removal, according to a news
item in Engineering News-Record (Grogan and Setzer 1992).
When the United States Army Corps of Engineers solicited bids
for debris removal immediately after Andrew, the Corps rejected
one-third of the bids as too high and awarded contracts bid at
approximately US $25 per cubic yard. One month later, the
awarded contracts had been bid at $7. One contractor charged with
debris removal received offers of heavy equipment that were
equally divided between “outrageous” and “acceptable.”
Although prices and wages can be understood to result from the
meeting of demand with supply, prices are set by individuals, and
local culture can play an important role in pricing decisions. In contrast to the Hurricane Andrew equipment example, where the drive
to maximize profit seems to have driven up costs, some contractors
and laborers elect to provide free or reduced price materials and
services, and still others may fix their costs at the preevent levels.
One of the writers (Olsen) observed this sense of altruism in interviews of local contractors after the U.S. Midwest floods of
June 2008.
Local, state, or national governments may enforce wage or price
controls, effectively setting an upper limit on reconstruction costs.
Antigouging regulations are fairly common. For example, after the
1995 Hurricane Marilyn in the U.S. Virgin Islands, the local
government widely publicized its antigouging position (Murphy
1995). Similarly, there may be trade restrictions on the movement
of materials and equipment between regions or on the licensing of
out-of-state contractors, effectively making supplies from outside
the affected area, if allowed, more expensive.
Reconstruction Timing
Regional factors may contribute to repair delays at an individual
property. A backlog of properties damaged in a previous event
or a construction boom in the region may delay work at a recently
damaged property, as suggested by Risk Management Solutions,
Inc., (2005) for repairs from the 2004 and 2005 Florida hurricane
seasons. The organization of the reconstruction effort, if any, may
determine the prioritization of work. Contractors may determine a
work schedule on a first-come-first-served basis, or they may
prioritize work according to potential profit. The government
may prioritize the repair of damaged properties, as was the case
after the 1999 Sydney hailstorm (Henri 1999). If the general population were evacuated, there may be a delay in reconstruction until
there is a critical mass of people to finance, permit, perform, and
inspect the repairs.
Debris removal can be critical before reconstruction begins at a
property. Storm surge in Hurricane Katrina pushed substantial
quantities of debris onto properties near the shore along the
Mississippi coast, hindering access (see, for example, Mosqueda
and Porter 2007). In addition to disposing of structural components
and building contents, there may be hazardous materials requiring
special attention, as was the case in several New Orleans hospitals
observed after Hurricane Katrina (Arendt and Hess 2006).
The efforts of local and national governments may also affect
the timing of reconstruction. Rebuilding may not begin until the
local government allows the release of building permits. For
example, as of June 30, 2008, in Cedar Rapids, Iowa—seventeen
days after floodwaters crested—the building department was issuing plumbing, electrical, mechanical, and building permits for flood
repairs outside the 500-year floodplain only (James Thatcher, personal communication, June 2008). The Cedar Rapids building department expected to release plumbing and electrical permits a few
days after this date, but they would not grant building permits until
the city decided on a reconstruction plan. For properties in the
100-year floodplain, the city had to consider the guidelines and regulations of the National Flood Insurance Program (NFIP). As of
June 30, it was not clear whether the city would allow any rebuilding on the 100-year floodplain or adopt the NFIP requirement that
properties with damage valued at more than 50% of the structure’s
value be rebuilt at least 0.3 m above the 100-year floodplain. The
city recommended that property owners clear debris, dry the structure, and wait for a city council decision. The approach taken in
Cedar Rapids suggests that local governments can affect the
progress of reconstruction. The municipality may actively influence repairs and rebuilding, or it may passively allow
reconstruction to be determined by property owners, financing,
and the availability of materials, labor, and equipment. Finally,
the building department may be overwhelmed, and permitting
and inspections may be delayed. The aftermath of the 2004 hurricane season in Florida is an example (Weintraub 2004/2005).
Contractor Fees
Typically in developed countries, construction contractors perform
the repairs. In such cases, contractors’ fees contribute to the costs of
reconstruction. In the aftermath of a large-scale natural disaster,
there may not be a competitive environment for contractor services.
Contractors typically increase overhead and profit as a buffer
against risk. Greater uncertainties in material and labor costs
and availabilities, as well as uncertain site access, will result in
greater repair costs. For example, the uncertainty in the weeks after
Hurricane Katrina affected reconstruction bidding: “While it is too
early to determine the full impact of the storm on construction
costs, the uncertainty it has stirred up undoubtedly will lead to
higher bids for projects just to cover the new level of risk” (Grogan
and Angelo 2005).
Contractors may change their bidding strategies or their rates for
overhead and profit margins. Some contractors may seek the highest margins, whereas others may not change their rates. The supply
of contractors may be affected by the local supply and by the
willingness and ability of outside contractors to travel to the
affected area.
Postcatastrophe increases in contractor profit margins are perhaps intuitive but difficult to document and quantify. One piece
of evidence is a Federal Emergency Management Agency review
of contract modifications awarded by the City of New Orleans to
inspect and repair storm drains after Hurricane Katrina. The city
allowed the contractor a 13% profit margin retroactively for work
that had been performed previously by subcontractors: work that
was contracted without competition, price analysis, or monitoring
of billing accuracy (Lankford 2006). In another post-Katrina example, the Federal Aviation Administration allowed nominal profits of
up to 15% for contractors providing response and recovery services, often without supporting evidence of subcontractor quotes,
possibly because “contract administration resources were stretched
quite thin and, understandably, … the top priority was to provide
transportation to deliver goods, equipment, and services as quickly
as possible to those in need of assistance.” A profit margin of 10%
is considered appropriate by several federal agencies, and FAA
NATURAL HAZARDS REVIEW © ASCE / MAY 2011 / 67
Downloaded 28 Jun 2011 to 198.11.27.29. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
acquisition policy and contract terms require subcontractor quotes
(Leng 2007).
General Economic Conditions
This paper has discussed issues specific to material prices, labor
wage rates, equipment costs, and contractor profit, but these
may in turn be affected by the general state of the economy. A predisaster construction boom may increase costs to rebuild damaged
properties because contractors are already in high demand. A lull in
construction increases the supply of available contractors, and thus
competition may keep costs low. Walker (2005) described two
examples:
After the Newcastle earthquake there was very little demand
surge because the building industry in New South Wales was
in recession, and there was sufficient surplus capacity of
building tradesmen and supplies available just down the road
in Sydney to maintain a competitive building environment.
The Canberra bushfires in January 2003 resulted in much less
damage than the Newcastle earthquake but they occurred during a building boom when there was no spare capacity in New
South Wales, with the result that the demand surge was very
significant—possibly of the order of 25 percent or more.
In addition to the local demand and supply of materials, labor,
equipment, and financing, the national and international supplies
may also influence the local costs.
The 2005 Hurricane Katrina provides an example of increased
costs resulting from the prevailing economic conditions. The
pre-Hurricane Katrina construction economy and experiences with
previous disasters affected expectations of construction cost increases. Before Katrina made landfall, there was an existing high
demand for construction materials and equipment (Hampton 2005).
The additional demand for reconstruction materials and the practice
of just-in-time supply led to expectations of price volatility and
shortages after Katrina (Nicholson 2006). Within one month of
the storm, a construction-cost consultant expected 10–15% higher
construction costs in the affected area, and one construction management firm increased estimates of material prices for commercial
projects by 8–10% and for residential projects by 10–15% (Grogan
and Angelo 2005). Cost estimators adjusted escalation factors in
bids in anticipation of demand surge (ENR Staff 2005).
Insurance Claims Handling and Decisions of an
Insurance Company
There are several ways that an insurer can handle the large volume
of claims after a large-scale natural disaster. An insurer may decide
not to verify claims under a certain value, as for example, after the
1999 Windstorms Lothar and Martin in France (Risk Management
Solutions, Inc. 2000). An insurer may have a preexisting agreement
with a contractor stipulating material and labor costs and a margin
for overhead and profit. Most commonly, an insurer sends a claims
adjuster to the property to identify the damage and estimate the
loss. For the same damage at, and policy conditions for, a property,
each of these insurance claims handling methods may result in a
different loss to the insurer.
Regarding the typical procedure of verifying the claim with a
claims adjuster, the characteristics of the adjuster or the methods
used may affect the loss as well. A well-trained and experienced
adjuster tends to adjust claims lower than an inexperienced adjuster
for the same damage (Kirk Beatty, personal communication, May
2009). Also, evaluating partial damage is more difficult than a total
loss. Some policies and events require the adjuster to distinguish
the cause of damage, and making these distinctions can require special expertise. A person unfamiliar with the type of damage may not
properly adjust the claim. For example, an adjuster from a seismically inactive area may be unable to distinguish foundation cracks
caused by settling versus those caused by an earthquake. (For that
matter, claims adjusters from seismically active California can have
the same problem.) An adjuster may not be able to distinguish damage caused by a previous event from normal wear and tear.
Claims adjusters anticipate materials costs and labor wage rates
by using price lists. Thomas (1976) described how U.S. insurers
handled catastrophe claims circa 1976: “As a result of their experience in numerous catastrophes since the 1930s the insurance companies have formulated catastrophe plans to better organize and
supervise the handling of claims.… One of the first steps taken
by the administrators of a catastrophe plan is to arrange for a
conference with local builders and contractors associations to discuss and agree when possible on unit costs for various building
items. These unit costs are published and then used as a guide
by the adjusters and builders in estimating the losses in the area.”
Currently, many adjusters use software to estimate reconstruction
costs. Xactware (2010) is the leading provider of pricing information for U.S. insurers, but it cautions the users of its price lists:
Xactware makes every effort to ensure pricing information …
represents market costs at the time of publication. Since actual
market prices can vary and may change rapidly, and since
many factors can affect the cost of a project (including—
but not limited to—labor, equipment, and material costs as
well as the rates and application of sales tax), we strongly
recommend customers monitor their local markets for any
such changes and adjust their estimate pricing as deemed
appropriate.
An adjuster unfamiliar with any pricing software may use it inappropriately (Postava 2008). The data contained in the software
may be for repair under ordinary conditions, and may be inappropriate for estimating losses in large-scale natural disasters. The
adjuster may not know how to properly account for rebuilding after
a large-scale natural disaster.
After natural disasters, insurance companies may make decisions that affect their loss. Insurers have faced pressure from several
sources in the aftermaths of past disasters, including their clients,
competition, and local and national politicians. The judiciary can
also influence insurers’ decisions through decided or anticipated
judgments on insurance disputes. An insurance company may
decide to pay for damage it deems excluded in its policy or not
enforce multiple deductibles in clustered events. These types of
payments may be at the discretion of the insurer and made to avoid
costly litigation or maintain a good public image. Depending on its
market segment and business model, an insurer may take great
effort to minimize the loss paid on each claim, or promptly pay
claims on its policies with little verification, or take a path between
these extremes. This issue of insurers’ decisions in the aftermath of
a large-scale natural disaster is on the border of what is and is not
demand surge. In this instance, the writers understand demand
surge to be the demand for reconstruction financing conflicting
with the limited supply from insurers. These decisions may affect
the limits to this supply.
Many of these insurance-related demand-surge issues were
documented after the 1906 San Francisco earthquake and fire.
The earthquake struck the San Francisco Bay area on April 18,
1906, and the last fire was extinguished four days later. Between
300 and 400 insurance claims adjusters, both local and from across
the United States, handled the large volume of claims on insurance
68 / NATURAL HAZARDS REVIEW © ASCE / MAY 2011
Downloaded 28 Jun 2011 to 198.11.27.29. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
policies. G. H. Marks of the London Assurance Corporation described the task of adjusting insurance claims after the earthquake
and fire (Marks n.d.). According to the standard procedure, claims
on insurance policies should be verified. However, the earthquake
and fire in San Francisco made this a nearly impossible task.
Complete destruction of the sites erased any evidence for the
adjusters; there were no neighbors or acquaintances of the claimant
who could verify the claim, nor could the original purchase be
verified by the retailer or wholesaler. Marks further noted that every
insurance company had lost most or all of their records in the fire.
Thus, the insurance companies could rarely verify that they even
covered the loss that was being claimed.
Some properties were indemnified by several companies. In the
months following the disaster, insurance companies tried to
develop a common approach to adjusting and settling these claims.
At a general meeting of the insurers on June 12 in Oakland,
California, across the Bay from San Francisco, the attendees voted
on a set of principles for adjusting claims. Essentially, the resolution stated that the insurance companies would apply a 25% reduction to the insured loss when the cause of damage—either
earthquake or fire—was unclear. (Standard insurance policies
excluded earthquake damage but covered fire damage. The insurers
estimated that 25% of the damage in San Francisco was caused by
the earthquake shaking and 75% by the fire.) Although otherwise
supporting the work of the General Adjusting Board, 35 companies
decided not to reduce payments to their policyholders. On June 21,
the (aptly named) Thirty-Five Companies adopted guidelines stating that claims on their policies should be settled on a case-by-case
basis according to the actual policy terms. In its concluding report
of December 31, 1906, the Committee of Five, which reviewed
claims on the policies of the Thirty-Five Companies, made recommendations to the companies on how to better handle claims after
future disasters (Hosford et al. 1906). The committee recognized
the unprecedented and difficult task faced by the insurance industry
after the 1906 San Francisco disaster. Nonetheless, the committee
suggested that insurance companies would have realized better and
less expensive results, in addition to “creating a better feeling
amongst policyholders,” if the companies had agreed to a common
approach much sooner. The committee reported that uncertainty
among policyholders made its work more difficult and resulted
in higher costs to the insurers.
In general, the committee found claimants to be “fair, patient,
and honest.” However, the committee acknowledged that, in their
haste to submit claims, policyholders often “innocently exaggerated their statements,” since the prevailing conditions made an accurate claim “almost impossible.” The committee noted two cases
in which “expert accountants” identified intentional fraud and reduced the total insurance payment by US$100,000, demonstrating
the value of the accountants’ work. If more widely employed, the
committee wondered, how many additional fraudulent claims
would have been identified?
In addition to issues under their control, insurers faced pressure
from their clients, competition, and the government. The General
Manager of Munich Re noted that, “In San Francisco we have experienced that companies which had stipulated in their policies the
exclusion of all direct and indirect earthquake losses were, due to
competition with other companies, obliged to pay their losses in
order not to lose their business for all future” (quoted in Hoffman
1928). This issue of payment by an insurer for excluded loss
may be an indirect result of the overwhelming demand for
reconstruction financing via insurance. If the scale of the 1906 disaster had been smaller, the insurance companies might not have
faced pressure from their policyholders or their competition to satisfy their clients beyond the terms of their policies.
U.S. courts ultimately ruled on disputes over fire following
earthquake exclusions in typical insurance policies. On October
5, 1908, the Ninth Circuit Court of Appeals upheld the finding
of a lower court that insurance companies were liable for losses
caused by fires indirectly ignited by the earthquake (Williamsburgh
City Fire Ins. Co. v. Willard 1908). The Court reasoned that, since
the insurers wrote their policies, any ambiguity in the policy language should favor the policyholder. A decision of the Second
Circuit Court of Appeals on November 13, 1911, confirmed payment to the insured in Norwich Union Fire Ins. Soc’y v. Stanton
(1911), using the same argument given in Williamsburgh City Fire
Ins. Co. v. Willard (1908).
Summary
The evidence for demand surge from historical events suggests four
important points regarding what demand surge is and is not.
1. Demand surge is not a new phenomenon. This paper has
offered evidence for demand surge in early 18th-century
England, the 1886 Charleston and 1906 San Francisco
Earthquakes, and the present day.
2. Demand surge is not limited to one region or country. There is
evidence for demand surge from the United States, the United
Kingdom, Australia, and continental Europe. Although evidence of demand surge is most readily available from these
territories, the apparent causes and mechanisms of demand
surge are not unique to them. Thus, there does not appear
to be strong reason to exclude the possibility of demand surge
in any region of the world.
3. Demand surge is not unique to one or two perils. Rather, it has
been observed following cyclones, earthquakes, floods, hailstorms, windstorms, and wild- or bush-fires (Olsen and Porter
2010). The extent of demand surge may be affected by the
peril, if a particular material or skill is in short supply.
However, the fundamental problem of supplying the demanded
services and goods is the same across perils.
4. Demand surge has many and varied causes that depend on the
particulars of each event. Since each large-scale natural disaster is unique, there is a unique explanation for demand surge in
each event. Nonetheless, themes emerge when considering
these events together, rather than the isolated incidents they
appear to be.
Conclusions
There are several inconsistent definitions of demand surge, none
of which is sufficiently precise to define a clear phenomenon.
“Demand surge” has been used to refer specifically to the increase
of materials prices and labor wages. Other sources expand this definition to include specific causes of higher costs, for example, construction and insurance services and the expansion of insurance
coverage. Still other sources understand demand surge to be the
difference between the expected (or modeled) loss versus the actual
loss. Without a standard, precise definition, there is confusion about
demand surge as a socioeconomic phenomenon, and there are various metrics of demand surge.
Historical events provide anecdotal evidence for demand surge.
In any particular natural disaster, there is a specific and unique explanation of demand surge. Considering many past large-scale
disasters, common themes emerge as plausible explanations for
why demand surge happens. This paper identified the following
themes: total amount of repair work; timing of reconstruction; costs
of materials, labor, and equipment; contractor overhead and profit;
NATURAL HAZARDS REVIEW © ASCE / MAY 2011 / 69
Downloaded 28 Jun 2011 to 198.11.27.29. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
general economic situation; insurance claims handling; and decisions of an insurance company. These themes are qualitative, proposed explanations for demand surge, and thus, they are also
hypotheses to be tested in future work. The development of these
themes provides a novel, mechanistic approach to the demand
surge problem.
Acknowledgments
This work was sponsored by Willis Group Holdings Limited
through the Willis Research Network. The writers appreciate the
support and interest of colleagues at Willis: Rowan Douglas,
Matthew Foote, Julie Serakos, Kyle Beatty, and Prasad Gunturi.
The assistance of the following individuals and institutions is also
gratefully acknowledged: Kathleen Tierney and Wanda Headley
of the Natural Hazards Center at the University of Colorado at
Boulder; Nicholas Butler and Kathleen Gray of the Charleston
Archive, Charleston County Public Library; Stephen Hoffius and
Susan Williams; Tim Wilson of Hawkeye Electric; Robert MuirWood, Chief Research Officer, Risk Management Solutions; David
Passman, Senior Vice President, Willis HRH; Kirk Beatty,
ServicePoint Specialized Processing Operation Director, Farmers
Insurance; and especially Auguste Boissonnade, Vice President
of Product Development, Risk Management Solutions.
The writers also thank two anonymous reviewers and Stéphane
Hallegatte for reviewing the manuscript and providing helpful comments.
References
AIR Worldwide Corporation. (2007). “AIR’s U.S. hurricane model update
incorporates enhanced methodology for estimating business interruption.” Press release, June 5.
Arendt, L. A., and Hess, D. B. (2006). “Hospital decision making in the
wake of Hurricane Katrina: The case of New Orleans.” MCEER-06SP01, MCEER, Univ. at Buffalo SUNY, Buffalo, NY.
Australian Securities and Investments Commission (ASIC). (2007).
“Making home insurance better.” Rep. 89, Canberra, Australia.
“Bricklayers in council.” (1886a). News and Courier (Charleston, SC),
Sep. 17.
Chock, G. Y. K. (2005). “Modeling of hurricane damage for Hawaii
residential construction.” J. Wind Eng. Ind. Aerodyn., 93, 603–622.
Defoe, D. (1704). The storm, Penguin Books, New York.
ENR Staff. (2005). “Rebuilding will keep pressure on costs.” Eng.
News-Rec., 255(12), 65.
Federal Alliance for Safe Homes and The Actuarial Foundation. (2006).
If disaster strikes will you be covered? Tallahassee, FL.
“Fixing the rate of wages.” (1886b). News and Courier (Charleston, SC),
Sep 7.
Florida International University. (2009). “Florida public hurricane loss
model.” Submitted to the Florida Commission on Hurricane Loss
Projection Methodology, Miami, FL.
Florida Statute. (2009). “Florida commission on hurricane loss projection
methodology; public records exemption; public meetings exemption.”
627.0628(3)(f), Tallahassee, FL.
Grogan, T., and Angelo, W. J. (2005). “Katrina keeps inflation roaring.”
Eng. News-Rec., 255(12), 66.
Grogan, T., and Setzer, S. W. (1992). “Hurricane triggers price storm.”
Eng. News-Rec., 229(13), 26.
Grossi, P., and Kunreuther, H., eds. (2005). Catastrophe modeling: A new
approach to managing risk, Springer, New York.
Grossi, P., and Muir-Wood, R. (2006). The 1906 San Francisco earthquake
and fire: Perspectives on a modern super cat, Risk Management
Solutions, Inc., Newark, CA.
Guy Carpenter & Company. (2005). “Model issues to consider in Katrina’s
wake.” Instrat Briefing, New York.
Hallegatte, S. (2008). “An adaptive regional input-output model and
its application to the assessment of the economic cost of Katrina.”
Risk Anal., 28(3), 779–799.
Hamblyn, R. (2005). “Introduction.” The storm, by D. Defoe, Penguin
Books, New York, x–xl.
Hampton, T. (2005). “Heavy demand for construction machines will push
up prices and slow deliveries.” Eng. News-Rec., 255(10), 32.
Henri, C. (1999). “The Sydney hailstorm: The insurance perspective.” Aust.
J. Emerg. Manage., 14(4), 16–18.
Hoffman, F. L. (1928). Earthquake hazards and insurance, The Spectator
Co., Chicago.
Hosford, A. R., Bament, W. N., Morrison, E. C., Corbet, J. C., and Seaman,
W. B. (1906). Report of the committee of five to the ‘thirty-five companies’ on the San Francisco conflagration, New York.
Howard, L. (2005). “Hurricane Katrina: A market-turning event?.”
Viewpoint, 30(2), 〈http://www.aaisonline.com/viewpoint/05fall2a
.html〉 (Jul. 30, 2009).
Huguenin, T., A. “The knights of labor.” (1886c). News and Courier
(Charleston, SC), Sep. 14.
Kuzak, D., and Larsen, T. (2005). “Use of catastrophe models in
insurance rate making.” Chapter 5, Catastrophe modeling: A new approach to managing risk, P. Grossi and H. Kunreuther, eds., Springer,
New York.
Lankford, J. (2006). “Interim review of hurricane Katrina activities, city of
New Orleans, Louisiana.” Office of Inspector General, U.S. Dept. of
Homeland Security, Washington, DC.
Leng, R. C. (2007). “Emergency transportation services contract: Lessons
learned from the 2005 Gulf Coast hurricanes.” Rep. No. FI-2007-030,
U.S. Department of Transportation, Washington, DC.
Marks, G. H. (n.d.). The San Francisco story—April 18–21, 1906, in the
Bancroft Library, Univ. of California, Berkeley, CA.
McAneney, J. (2007). “Post-event claims inflation (PECI).”
Quarterly Newsletter 7, Risk Frontiers, Macquarie Univ., Sydney,
Australia.
Mosqueda, G., and Porter, K. (2007). “Engineering and organizational issues before, during, and after hurricane Katrina, damage to engineered
buildings and lifelines from wind, storm surge and debris in the wake of
hurricane Katrina.” MCEER-07-SP03, MCEER, Univ. at Buffalo
SUNY, Buffalo, NY, 54.
Munich Re Group. (2006). “Annual review: Natural catastrophes 2005.”
Knowledge series: Topics geo, Munich, Germany.
Murphy, C. (1995). “Boats ferrying supplies, good will.” St. Petersburg
(FL) Times, Sep. 21, 8A.
Nebraska Administrative Code. (2009). “Standards for prompt, fair, and
equitable settlements applicable to fire and extended coverage
type policies.” Title 210, chapter 60, section 010.01, subsection (B),
Lincoln, NE.
Nicholson, T. (2006). “Big home centers’ influence sways the broader materials market; smaller contractors are starting to bite, but big fish are
elusive.” Eng. News-Rec., 256(11), 29.
Norwich Union Fire Ins. Society v. Stanton et al. 191 F. 813 (2d Cir. 1911).
Okuyama, Y. (2007). “Economic modeling for disaster impact analysis:
Past, present, and future.” Econ. Syst. Res., 19(2), 115–124.
Olsen, A. H., and Porter, K. A. (2010). “What we know about demand
surge.” Technical Rep. SESM-10-1, Dept. of Civil, Environmental,
and Architectural Engineering, Univ. of Colorado at Boulder.
Postava, J. A. (2008). “Common estimating errors.” 〈www.simsol.com/
PDF%20Documents/Common_ Estimating_Errors.pdf〉. (Apr. 15,
2009).
Risk Management Solutions, Inc. (2000). “RMS analyses indicate majority
of French reinsurance programs blown by recent windstorms.” Press
release, Jan 7.
Risk Management Solutions, Inc. (2005). “Repair costs in Southeastern
U.S. remain 20% to 40% above average as 2005 hurricane season begins.” Press release, Jun. 1.
Rose, A. (2004). “Economic principles, issues, and research priorities in
hazard loss estimation.” Chapter 2, Modeling spatial and economic impacts of disasters, Y. Okuyama and S. E. Chang, eds., Springer-Verlag,
New York, 13–36.
Ruquet, M. E. (2009). “Demand surge not driven by economic demand.”
70 / NATURAL HAZARDS REVIEW © ASCE / MAY 2011
Downloaded 28 Jun 2011 to 198.11.27.29. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
〈http://www.propertycasualty360.com/2009/03/20/demand-surge-notdriven-by-economic-demand〉 (Mar. 19, 2011).
Sirkin, A. (1995). “Engineering overview of Hurricane Andrew in South
Florida.” J. Urban Plann. Dev., 121(1), 1–10.
Stavro, B. (1998). “AIG says it will increase its stake in 20th Century.”
Los Angeles Times, Apr. 11, D1.
Subcommittee on Ratemaking of the Casualty Committee. (2000). “Treatment of catastrophe losses in property/casualty insurance ratemaking.”
Actuarial Standard of Practice 39, Doc. No. 072, Actuarial Standards
Board, Washington, DC.
Sweetman, K., and Morris, R. (1999). “12 vultures: Tradesmen profiteer
from hail misery.” The Daily Telegraph, (Sydney, Australia), Apr. 17,
Local 1.
Thomas, P. I. (1976). How to estimate building losses and construction
costs, 3rd Ed., Prentice-Hall, Englewood Cliffs, NJ.
Walker, G. (2005). “Exploring the benefits and overcoming the challenges
in insurance risk modelling.” CPA Australia Insurance Conf., Aon,
Queensland, Australia.
Weintraub, L. A. (2004/2005). “Governor Bush’s emergency orders creates
opportunities for contractors and consumers.” South East Florida
Constructor, Winter, 16–17.
Williamsburgh City Fire Ins. Co. v. Willard. 104 F. 404 (9th Cir. 1908).
Xactware (2010). “January 2010 price list update now available.” Xactware
News from 2010, 〈http://www.xactware.com/news/release20100101〉
(Mar. 14, 2011).
NATURAL HAZARDS REVIEW © ASCE / MAY 2011 / 71
Downloaded 28 Jun 2011 to 198.11.27.29. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org